A simple approach to robot navigation based on cooperative neural networks

摘要

This paper presents a new neural navigation and obstacle avoidance model based on multiple neural networks cooperation. In the proposed system, a combination of a Self-Organising Map and Multilayer Perceptrons makes use of local information to reach a final objective into a small maze, requiring neither supervision nor previous knowledge. The use of combined artificial neural network architectures in the domain of autonomous agent navigation provides promising results.